ABSTRACT
With increasing circuit densities and correspondingly finer feature sizes, the design and fabrication of the next generation of computer devices and their packages are becoming more complex and difficult. Problems of increased sensitivity to minute amounts of contamination and subtle phenomena such as electromigration are well known and require the utmost attention of the device or package designer. However, a more treacherous problem lies in assessing the thermal-mechanical stability of any particular wiring design and a choice of materials. The dual specters of delamination or cracking are ever present and often obscured by complexities of the process flow. In
Figure 1 shows an array of copper lines embedded in a polyimide insulator. A dense array of copper lines embedded in a polyimide insulator is idealized by an infinite array of parallel lines. Periodic boundary conditions can be applied to the infinite array so that only a small repeat unit needs to be considered. A thin layer of silicon nitride separates the top of the copper lines from the overlying polyimide. A similar layer of silicon nitride lies at the bottom of the trenches formed by the copper lines. The object is to anticipate what failure modes could arise from the thermal expansion mismatch between the copper and polyimide. If one is forced to consider the universe of all feasible mechanisms, this problem is for all practical purposes beyond our abilities. However, relying on past experience, the problem can be whittled down by making a number of appropriate assumptions: (1) The failure is assumed to occur as a delamination.
